GIGABYTE's Entire 6 Series Ready to Support Native PCIe Gen. 3

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
GIGABYTE TECHNOLOGY, a leading manufacturer of motherboards, graphics cards and computing hardware solutions today announced their entire range of 6 series motherboards are ready to support the next generation Intel 22nm CPUs (LGA1155 Socket) as well as offer native support for PCI Express Gen. 3 technology, delivering maximum data bandwidth for future discrete graphics cards.

Wanting to provide maximum upgradeability to customers, GIGABYTE has enabled native support for PCI Express Gen. 3 across the entire range of GIGABYTE 6 series motherboards, including the recently launched G1.Sniper 2 motherboard, when paired with Intel’s next generation 22nm CPUs. By installing the latest BIOS for their 6 series motherboards today, users can be assured they are ready to take advantage of all the performance enhancements tomorrow’s technologies have to offer.
 
I don't see graphics card needing to push 16gb/s for a while, since most people use 1 HD monitor with 1080p right now and it's just fine.

most of the push from using pcie3.0 is from enterprise and storage side, where drive controllers are connected to HDDs using pcie3.0.
 
I think its a good venue for an upgrade path. Sandybridge --> Ivybridge without a must have of getting a new MB.

I mean unless theres some new standard that is due to come out before Ivy....
 
That makes me really happy. I really like my GA-Z68X-UD3H-B3 board and would like to keep it for a while. I couldn't care less about PCIe Gen 3 (I run onboard graphics) but I love the fact they confirmed that I will be able to throw a Ivybridge in it.
 
I don't see graphics card needing to push 16gb/s for a while, since most people use 1 HD monitor with 1080p right now and it's just fine.

most of the push from using pcie3.0 is from enterprise and storage side, where drive controllers are connected to HDDs using pcie3.0.

16 lanes of pci-e 3.0 is actually 128gb/s, 16GB/s
 
So, now, is the only thing the new chipsets will have is triple channel RAM that P67 users don't already have?
 
so if gigabyte is able to enable pci-e 3.0 support with just a bios update does that mean that other manufactures are able to also and are just holding out ?
 
SO after doing some research it appears that everyone says somthing different.

According to gigabyte you only need a bios update because the physical traces for pcie 2.0 and 3.0 are identical.

According to MSI and ASROCK you need different pcie controler chips.

And according to Asus the PCIE controller chips that MSI and ASROCK are touting are untested and have no guarantee of actually working and we may only need a bios update but we may need entirely different boards and further more ASUS states that any board using a NF200 chip will proably not be pcie 3.0 capable.

http://forums.anandtech.com/showthread.php?p=32115449

http://translate.google.com.tw/trans...0.html&act=url
 
SO after doing some research it appears that everyone says somthing different.

According to gigabyte you only need a bios update because the physical traces for pcie 2.0 and 3.0 are identical.

According to MSI and ASROCK you need different pcie controler chips.

And according to Asus the PCIE controller chips that MSI and ASROCK are touting are untested and have no guarantee of actually working and we may only need a bios update but we may need entirely different boards and further more ASUS states that any board using a NF200 chip will proably not be pcie 3.0 capable.

http://forums.anandtech.com/showthread.php?p=32115449

http://translate.google.com.tw/trans...0.html&act=url

It would depend on which PCI-E slots you are talking about. When you put in an Ivy Bridge CPU you are putting in a new PCI-E controller for the GPU slots. The board only has the traces that go from the CPU socket to the slot. The PCI-E slots that run off of the southbridge will need an upgraded controller on the board to be PCI-E Gen. 3.
 
It would depend on which PCI-E slots you are talking about. When you put in an Ivy Bridge CPU you are putting in a new PCI-E controller for the GPU slots. The board only has the traces that go from the CPU socket to the slot. The PCI-E slots that run off of the southbridge will need an upgraded controller on the board to be PCI-E Gen. 3.

also, the switching bridge chips that enable one x16 slot to become two x8 slots, needs to be PCIe 3.0 complaint. I dunno if the main x16 slot would be affected by the switch chips, however...
 
Sure would be nice if somone from gigabyte would come in here and tell us the whole story about this.

My personal guess is that Asus's take on the issue seems the most likely. I.E yes you can get pcie 3.0 capability on an existing board with just a bios update but only at pcie 3.0 x8 speeds which would make sense as PCIE 3.0 x8 would have the same bandwith as PCIE 2.0 x16 (+ the 20% gained from the improved communication protocols). or in other words it's all just a marketing gimmick.

That being said if someone from gigabyt would actually be willing to gaurante real/full pcie 3.0 speeds that would certainly sway my buying decision for the build im working on.
 
There is lots of nonsense being talked about here, and some people do not FOCUS on the important stuff.

1) @ Brom42 : Of course we are not talking about the pci-express slot coming from the mobo chipset...DUH! even the supposedly already pci-e 3.0 comppliant motherboards from asrock/msi do not claim 3.0 support on those slots.

2) @ Kinjo : No you have understood it totally wrong, it does NOT make sense to have pci-e 3.0 @8x because it is the same bandwidth as pci-e 2.0 @16x.

The controller is in the cpu, and the physical interconnects lead to the pci-e slot (for simplicity's sake lets assume a motherboard with only 1 pci-e x16 slot). Now, since physically the pci-e 2.0 and 3.0 are exactly the same and won't need different interconnects and are the exact same pins, the only thing that matters is if Ivy Bridge has the same internal structure as sandy bridge. Meaning, if the existing configuration that connects the slot pins to a certain part of the cpu socket , if that part will house the pci-e controller on the Ivy bridge too.
Assuming intel is nice enough and considering that if this didn't happen it would be a total joke for the manufactureres already touting support for 3.0 , we should take this for granted.

Now onto the bandwidth part. Since the physical configuration is correct, the bandwidth is all about the controller. If the controller is pci-e 3.0 in the cpu you will get full 3.0 speeds, not 8x or any other mumbo jumbo stuff. Its the way the controller treats the physical part(aka type of signal, MHz etc) that enables the bandwidth. It's not like the pins in current motherboards pose a limit to 8gb/s so the 3.0 controller will be limited to 8gb/s=8x speed... Think of it like the sata 1/2/3 that have the same 7 pins and you can use the same cables (even though some marketeers state sata3 support and stuff, its bullcrap) or like ADSL/VDSL etc where on the same cables, changing the controller you change the way the signal is transfered and thus the capacity of it.

So forget the 8x stuff, if it works, it will work 16x.

Now what Asus is saying and does indeed make sense, is that if a motherboard has the ability to split the 16x lane to 2 lanes of 8x, then that splitter might pose a problem, because there you add an extra part to the chain that is electrical and not just a physical junction. It's like this hub renews the incoming signal and splits it or doesnt before it goes to the slots, so if the hub cant handle the pci-e 3.0 signal you got problems.

Now I see 3 different scenarios here depending on the way this splitter works :

1) It leaves the signal unaffected and both slots operate as plannes at 8x or 1 slot at 16x.
This is the best scenario but I find it very unlikely. It's sensible that the hub receives full 16x 3.0 bandwidth, but then creates 2 lanes of 8x 2.0 speed. This is what is is supposed to do unfortunately. It doesnt matter that it receives 16gb/s , it is programmed to offer 2 slots of 4gb/s each. Think of it like a tank with 1 intake and 2 outtakes. The 2 outtakes have 4cm diameter each and the intake is 8cm diameter. Now whether you change the 8cm diameter to 16cm or not, unfortunately the outtakes are fixed at 4cm each and you can do nothing about that when you only have the ability to change the intake (=cpu).

2) A pretty good scenario that IS possible, is that when the cinfiguration requires for only 1 16x slot to work, then the splitter is TOTALLY BYPASSED. If this is the way it works, then at least those motherboards will take the 3.0 pci-e on the first slot and work as planned, just losing the ability to do SLI/Xfire.

3)The splitter is always in use and either provides 1 outtake of 16x 2.0 or 2 outtakes of 8x 2.0 . See the scenario explained in 1, its like the splitter has the ability to change the outtakes but only in its spesified configurations, and a changing the intake can do nothing about that.
Thats the worst scenario as those motherboards wont even get 1 working 3.0 slot.

Now which of the 3 will happen, depends on how the splitter works, and this only an Asus/other engineer can tell us about.
I would assume the second scenario is the most probable and this is kind of good news. I would trade sli/xfire capability for a full fledged pci-e 3.0 slot anyday and I am sure most would too.

Ironically, if the third scenario is what ends up happening, then the low end boards will get support and the high-end wont. IE , the P8Z68 LE will get pci-e 3.0 and all the other P8Z68 boards wont...

Sorry for the huge post, jsut trying to clear up some confusions you got, and give an inventive to some engineer to tell us actually how this splitter chip works, so we can estimate what will happen when Ivy bridge comes out....
 
Back
Top